Project file verification and correction method, system and equipment based on combination of project full life cycle and table

Abstract

The invention relates to a project file verification and correction method, system and equipment based on a project full life cycle combined table, and relates to the technical field of project management.The method comprises the steps that in an e report platform, the table is established on the basis of all data of a project file for establishing the project full life cycle, the table is automatically translated into training data, and the training data are sent to a server; the training set is convenient to collect. The method comprises the following steps: training a deep semantic model by utilizing a training set, performing file verification and classification processing on an uploaded project file through the model, performing comparison based on a file verification result and a table, determining whether the project file is wrong according to a comparison result, and outputting a reminding category for the wrong project file. According to the technical scheme, automatic project file verification and correction and automatic arrangement of the project files are achieved, the file management difficulty is remarkably reduced, the overall efficiency is improved, and the technical problems that in the prior art, due to the fact that automatic project file verification and correction management cannot be achieved, the efficiency is low, and the management difficulty is large are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of project management, and in particular to a project archive verification and correction method, system and device based on a project full life cycle and a table. BACKGROUND

[0002] At present, the collection and archiving of power project archives usually requires manual input and uploading, and it is impossible to view the archive data from a macro project perspective. In the project file archiving work, due to the large number of files and disordered classification, archive missing may occur at each project node, and it is impossible to see from the project full life cycle whether the project archive is missing or has errors. For example, manual uploading of files is prone to errors such as incorrect file location.

[0003] For the errors and omissions, reasonable verification and correction is needed. The existing project archive management scheme still relies on manual verification and correction, such as manually verifying the location, classification and information of project files, and manually correcting errors such as incorrect file location. With a large number of project files and multiple classifications, in the absence of automatic sorting and error correction, relying on manual correction will increase the difficulty of project archive management and reduce overall efficiency. SUMMARY

[0004] The present application provides a project archive verification and correction method, system and device based on a project full life cycle and a table, which automatically sorts and corrects project archives based on a project full life cycle and a table. Specifically, for a large number of project files collected by an e-report platform, a table is established to facilitate the collection of a training set for training a deep semantic model. The model is used to verify and classify uploaded project files, realize automatic project archive verification and correction, and automatically sort project files, thereby significantly reducing the difficulty of archive management and improving overall efficiency.

[0005] In a first aspect, the present application provides a project archive verification and correction method based on a project full life cycle and a table, comprising:

[0006] In the established adaptive e-report platform, taking each project archive of the project full life cycle as a benchmark, a table is established based on historical complete data of different project types;

[0007] Labels and training data are extracted from the table to form a training data set;

[0008] A deep semantic model is established, the training data set is used to train the deep semantic model, and a trained deep semantic model is obtained;

[0009] The uploaded project files are verified and classified using a trained deep semantic model. The file verification results are then compared with the categories in the table to obtain the comparison results.

[0010] Based on the comparison results, a correction reminder message is output.

[0011] Optionally, create tables based on the complete historical upload data for different project types, including:

[0012] Identify the archived data column and attachment column from the complete historical upload data of different project types;

[0013] Convert the archived data column into tag data and tag name data, and extract the attachment name data from the attachment column;

[0014] Create a table, and update the label column in the table based on the label data, update the label name column in the table based on the label name data, and update the data name column in the table based on the attachment name data, respectively;

[0015] In the table, a mapping relationship is established between each data in the label column, each data in the label name column, and each data in the data name column, and the table is updated through the mapping relationship;

[0016] In this table, each data point in the label column serves as a label, each data point in the data name column serves as training data, and each data point in the label name column serves as a label name used for correction reminders.

[0017] Optionally, updating the label column in the table based on the label data includes:

[0018] The tag data is converted according to a preset conversion rule to obtain the target tag;

[0019] Update the label column in the table based on the target label.

[0020] Optionally, the uploaded project files are processed for file verification and classification using a trained deep semantic model. The file verification results are then compared with the categories in the table to obtain the comparison results, including:

[0021] The uploaded project files are processed by a trained deep semantic model to perform file verification and classification, and the file verification results include the category labels of the uploaded project files output by the deep semantic model.

[0022] For the uploaded project file, the category tag is compared with the tag corresponding to the uploaded project file in the table to obtain the comparison result.

[0023] Optionally, the uploaded project files can be processed for file verification and classification using a trained deep semantic model to obtain file verification results, including:

[0024] In the deep semantic model, the uploaded project file is identified and extracted based on the input to obtain the file name;

[0025] The file name is processed by word segmentation mapping to construct vector information, which includes BERT word vectors.

[0026] Based on the vector information, feature analysis is performed to obtain text features;

[0027] The text features are used for classification and verification to obtain the file verification result.

[0028] Optionally, the file name is processed through word segmentation mapping to construct vector information, including:

[0029] The file name is processed by word segmentation mapping to obtain a word list, which includes at least one text word extracted from the file name;

[0030] Based on the text segmentation described above, an index mapping process is performed to establish an index list;

[0031] Based on the index list, BERT inference is performed to obtain vector information.

[0032] Optionally, based on the comparison results, correction reminder information is output, including:

[0033] When the comparison result is inconsistent, obtain the first label name of the category label in the table;

[0034] Get the name of the second tag of the uploaded project file in the table;

[0035] Based on the first tag name and the second tag name, a correction reminder message is generated and output.

[0036] Optionally, after creating the table, the following may also be included:

[0037] Extract the target attachment name and category from the download request of the target object;

[0038] Based on the target attachment name and the category content, the table is traversed, and a table organization method is used to perform hierarchical retrieval to obtain the file data to be downloaded;

[0039] Based on the data of the files to be downloaded, a compressed package is constructed by recursively adding files;

[0040] Download the compressed package using the preset download tool.

[0041] Secondly, this application provides a project file verification and correction system based on the entire project lifecycle and combined with tables, including:

[0042] The table creation module is used to create tables in the established adaptive e-reporting platform, based on the project archives throughout the entire project lifecycle and the complete historical data uploaded for different project types.

[0043] The training dataset extraction module is used to extract labels and training data from the table to form a training dataset.

[0044] The model training module is used to build a deep semantic model, and to train the deep semantic model using the training dataset to obtain a trained deep semantic model.

[0045] The verification and correction module is used to perform file verification and classification processing on uploaded project files through a trained deep semantic model, and to compare the file verification results with the categories in the table to obtain the comparison results.

[0046] The correction reminder module is used to output correction reminder information based on the comparison results.

[0047] Thirdly, this application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0048] Memory, used to store computer programs;

[0049] When a processor executes a program stored in memory, it implements the steps of the project file verification and correction method based on a table combining the entire project lifecycle as described in any of the first aspects.

[0050] In summary, this application's embodiment innovatively establishes tables to facilitate training set collection and trains a model for classifying and verifying archival files. Specifically, in the e-report platform, tables are created based on the data from project archives throughout the entire project lifecycle. These tables are automatically translated into training data, facilitating training set collection. A deep semantic model is trained using the training set. The model performs file verification and classification on uploaded project files. The verification results are compared with the tables to determine if any project files are erroneous. No action is taken on error-free project files; for erroneous files, a reminder message is output. This achieves automated project archive verification, correction, and organization, significantly reducing the difficulty of archive management, improving overall work efficiency, and effectively solving the technical problems of low management efficiency and high management difficulty caused by the inability of existing technologies to achieve automated project file verification and correction management. Attached Figure Description

[0051] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0052] Figure 1 A flowchart illustrating a project file verification and correction method based on a table and the entire project lifecycle, provided for embodiments of this application;

[0053] Figure 2 This application provides an example diagram illustrating the automated collection of project data based on tables on the eReport platform.

[0054] Figure 3 This is a flowchart illustrating the steps of an optional embodiment of the present application for a project file verification and correction method based on a table-based approach throughout the project lifecycle.

[0055] Figure 4 This is an example diagram of automatic translation of a table-based training set provided in this application;

[0056] Figure 5 This is a one-click download example provided in this application. Figure 1 ;

[0057] Figure 6 This is a one-click download example provided in this application. Figure 2 ;

[0058] Figure 7 This is a flowchart of a deep semantic model performing classification prediction, provided as an example in this application;

[0059] Figure 8 A structural block diagram of a project file verification and correction system based on a table and the entire project lifecycle, provided in this application embodiment;

[0060] Figure 9 This is an example diagram of an electronic device provided as an optional example of this application. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] To facilitate understanding of the embodiments of this application, further explanations and descriptions will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0063] Figure 1 This application provides a schematic diagram of project file verification and correction based on a table combined with the entire project lifecycle. The method may specifically include the following steps:

[0064] Step 110: In the established adaptive e-reporting platform, based on the establishment of project archives for the entire project lifecycle, create tables based on the historical uploaded complete data of different project types.

[0065] In the developed e-reporting platform, for each project file that has established a complete project lifecycle, tables are created based on the project type of each file and utilizing the historical uploaded complete data. This historical uploaded complete data may include, but is not limited to, project number, acceptance details, archived materials, and attachments. The tables can be used to extract training datasets; therefore, the tables mainly include three columns: a label column, a label name column, and a data name column. The label name column serves as the table category, indicating the type of each attachment.

[0066] Taking power projects as an example, refer to Figure 2 As shown, in the e-reporting platform, the acceptance content is the implementation progress of the power project. Archived materials are the materials required for the current progress. Files are named according to certain preset standards and updated to the table, such as the first file of progress 1 (1-1xxx), the second file of progress 1 (1-2xxx), the first file of progress 2 (2-1xxx), the second file of progress 2 (2-2xxx), etc. Attachments are where files are uploaded.

[0067] For example, the project archives that have been established for the entire project lifecycle include a directory of the entire project lifecycle, i.e., project process nodes. Among them, a power project may include preparation phase 1, agreement phase 2, etc. The implementation phases of different power projects can be used as directories, and the corresponding project files can be set in the directories.

[0068] Step 120: Extract labels and training data from the table to form a training dataset.

[0069] Step 130: Establish a deep semantic model and train the deep semantic model using the training dataset to obtain a trained deep semantic model.

[0070] Step 140: Perform file verification and classification processing on the uploaded project files using the trained deep semantic model, and compare the file verification results with the categories in the table to obtain the comparison results.

[0071] Step 150: Based on the comparison results, output correction reminder information.

[0072] The following is a unified description of steps 120-150 above:

[0073] In related technologies, power project archives are numerous and disorganized. Specifically, with business development, the number of archives has exploded, resulting in a massive accumulation of documents. Furthermore, the document classification standards are unclear and inconsistent; different departments or personnel have varying understandings and practices regarding document classification, leading to a chaotic and disorganized system. This makes finding specific documents akin to searching for a needle in a haystack, consuming significant time and effort and severely impacting the efficiency and quality of archive management.

[0074] To address the technical problems of the existing technologies mentioned above, this embodiment utilizes a deep semantic model to automatically verify archived and uploaded project files. Specifically, based on the constructed table, this embodiment can quickly translate the training dataset, i.e., the training set, from the table, avoiding the cumbersome training set collection process and facilitating model training. The data names in the data name column are extracted from the table as training data, and the labels corresponding to the data names are extracted from the table as labels for the training data. These labels are then mapped to the training data to obtain the training set. For example, if the label column in the table contains label 11, and the data name column contains the data name XXX project proposal document, then label 11 is mapped to XXX project proposal document, forming the training set.

[0075] Then, a deep semantic model (also known as a deep semantic learning text classification model) is constructed. The training set is input into the deep semantic model for model training until the model training is completed, and a trained deep semantic model is obtained.

[0076] When a new project is uploaded and archived, the project files of that project can be used as the uploaded project files. At this time, the table can be updated with data for the uploaded project files, including updating the tags, tag names, and attachment names corresponding to the uploaded project files in the table.

[0077] The uploaded project file is classified and verified using a trained deep semantic model. The category to which the uploaded project file belongs is identified, and the category label is output as the file verification result.

[0078] Then, for the uploaded project file, the corresponding tags are extracted from the table and compared with the category tags to determine whether the tags of the uploaded project file in the table are consistent with the category tags output by the model, and the comparison results are obtained.

[0079] Finally, a correction reminder is output based on the comparison results. For example, when the comparison is inconsistent, that is, when the surface model's verification classification of the project file is inconsistent with the existing category of the project file in the table, a correction reminder can be output; when the comparison is consistent, no correction reminder needs to be output.

[0080] Therefore, this embodiment utilizes existing tables to automatically translate them into training data, facilitating the collection of training sets and thus improving model training efficiency. Finally, the model is used to automatically verify and correct archived project files, effectively improving the efficiency and quality of file management.

[0081] As can be seen, in this application embodiment, based on the data of the project archive established throughout the entire project lifecycle, a table is created in the e-report platform. The table is automatically translated into training data, facilitating the collection of the training set. A deep semantic model is trained using the training set. The model performs file verification and classification on uploaded project files. Based on the file verification results and the table, the system determines whether the project files are erroneous. For erroneous project files, a reminder category is output. This achieves automated project archive verification and correction, as well as automatic organization of project files, significantly reducing the difficulty of archive management, improving overall efficiency, and solving the technical problems of low efficiency and high management difficulty caused by the inability to achieve automated project file verification and correction management in existing technologies.

[0082] Reference Figure 3 The diagram illustrates a step-by-step flowchart of a project file verification and correction method based on a table and the entire project lifecycle, according to an optional embodiment of this application. The method may specifically include the following steps:

[0083] Step 310: Based on the establishment of project archives for each project throughout its entire lifecycle, create tables using historically uploaded complete data for different project types.

[0084] Optionally, in this embodiment of the application, a table is established based on historical uploaded complete data of different project types. This may include: identifying archived data columns and attachment columns from the historical uploaded complete data of different project types; converting the archived data columns into tag data and tag name data, and extracting attachment name data from the attachment columns; establishing a table to update the tag columns in the table based on the tag data, the tag name columns in the table based on the tag name data, and the data name columns in the table based on the attachment name data; establishing a mapping relationship between each data in the tag columns, each data in the tag name columns, and each data in the data name columns in the table, and updating the table through the mapping relationship; wherein, each data in the tag columns in the table serves as a tag, each data in the data name columns serves as training data, and each data in the tag name columns serves as a tag name for correction reminders.

[0085] Based on the historical uploaded complete data of different power project types (such as technical renovation projects, science and technology projects, and repair projects), archived data and attachment data columns are extracted. The data in the archived data column is processed according to certain rules to be converted into tags and tag names. The file names of attachments are extracted from the attachment column as the attachment name data. For blank data in the attachment column, their corresponding tags and tag names are automatically deleted.

[0086] Next, create a table. In the corresponding label column, add the converted labels to the label column and update the table. In the corresponding label name column, add the converted label names to the label name column and update the table. In the corresponding training data column, add the converted attachment name data to the training data column and update the table. When updating the table, since the label data, label name data, and training data are all extracted from historically uploaded complete data, a mapping relationship can be established between these three types of data, and corresponding processing can be performed.

[0087] For example, refer to Figure 4 As shown, Figure 4 The displayed content is the complete historical upload data (and) Figure 2 Correspondingly, in the archived data of the historical uploaded complete data, there is an example data in the column of archived materials: "1-1 Project Application Document". In the attachment column, the attachment name is "XXX Project Establishment Document". Then, "1-1" and "Project Application Document" can be extracted from the archived materials column (which can be extracted by matching through regular expressions, etc.). "1-1" is converted to "11" as a label, and "Project Application Document" is used as the label name; the attachment name "XXX Project Establishment Document" is used as training data.

[0088] Similarly, based on the logic of the example above, 2-6 are converted to 26, and the file names are extracted from the corresponding attachment column, such as "xxx project acceptance report," "xxx contract," etc. At the same time, blank attachment data and their corresponding labels are automatically deleted, forming the training dataset. The labels and training data in the table constitute the training dataset.

[0089] For example, the extracted archive data and attachment data columns can be seen in Table 1 below:

[0090]

[0091] Table 1

[0092] The data in the table can be understood as shown in Table 2 below:

[0093]

[0094] Table 2

[0095] Therefore, this embodiment uses existing tables to automatically translate them into training data, which facilitates training set collection and enables efficient training set acquisition and model training.

[0096] In one optional embodiment, updating the label column in the table based on the label data may specifically include: converting the label data according to a preset conversion rule to obtain a target label; and updating the label column in the table based on the target label.

[0097] In practice, the format of the labeled data may not conform to the requirements of the label. Therefore, a preset transformation rule can be used to convert the labeled data into the label required by the training dataset. For example, the label data "1-1" contains hyphens, which does not meet the label requirements. Therefore, the transformation rule is used to remove the hyphens, resulting in the label "11".

[0098] In an optional embodiment, after establishing the table, this embodiment may further include: extracting the target attachment name and category content from the download request of the target object; traversing the table based on the target attachment name and the category content, performing hierarchical retrieval using a table organization method to obtain the file data to be downloaded; recursively adding based on the file data to be downloaded to construct a compressed package; and downloading the compressed package using a preset download tool.

[0099] In related technologies, existing technologies have failed to form a standardized project archiving process. In addition to the fact that errors are prone to occur when uploading project files, there are also the following problems: 1. Cumbersome download: It is difficult to download the required files with one click; 2. Cumbersome approval process: The approval of archived materials involves multiple links and departments. The existing approval methods are mostly paper-based circulation or decentralized online approval. The process is complex and opaque, making it difficult to achieve efficient collaboration and easily leading to approval delays or errors.

[0100] In this regard, this embodiment utilizes a table to enable one-click download of the required project files and trial procedures, thereby improving overall work efficiency.

[0101] In its implementation, the project file verification and correction method based on this embodiment can be used to build a management system. This system comprehensively covers multiple core modules such as document collection, approval workflow, and one-click download and automatic transfer. Its core objective is to significantly improve the standardization and overall efficiency of file archiving. Considering the traditional archiving process, where project files are numerous and poorly categorized, each collection requires individual document approval, which is not only inefficient but also prone to errors and omissions, posing a significant challenge to management. The management system in this embodiment utilizes an e-reporting platform to achieve end-to-end digital management from collection and approval to archiving.

[0102] Specifically, after a user uploads project attachments, the system can verify whether the file name corresponds to the uploaded form, thereby ensuring the integrity and compliance of the archived materials and greatly improving the accuracy and efficiency of the approval process.

[0103] To address the challenges of managing archived files, the system innovatively developed a one-click download function. This function not only allows users to download required files with a single click but also features automatic categorization and storage capabilities. Users can automatically categorize and store required files based on the type of materials from the acceptance phase or project archives. The code works as follows: First, based on different module names, the files associated with each module are categorized. After categorization, each module is converted into the specific format required by the code.

[0104] Reference Figure 5 and Figure 6 As shown, after logging into the system, the user (i.e., the target object) enters relevant information as needed. For example, the relevant information includes, but is not limited to: ① Attachment Name: Attachment; ② Module Category Content: Archived Materials. The user clicks Submit.

[0105] Then, the system can automatically organize files according to their naming conventions and download compressed packages, such as... Figure 7 As shown, folders will not appear if there are no attachments. Specifically, the system iterates through all attachment columns, checking if data exists. If it does, the file data is retrieved and organized; if not, such as when the relevant field value is empty, the system searches upwards to complete the data. This is compatible with situations where customers only fill in the first field of the same type of data.

[0106] Finally, files are recursively added to the compressed package according to the hierarchical relationship. Once the compressed package is built, it can be downloaded using a browser or other download tools.

[0107] In addition, this system can also be used to automate workflow approvals.

[0108] In one alternative implementation, after the form is constructed, a workflow review can be performed. During the approval process, approval nodes can be established through a built-in workflow engine, and approval forms can be constructed based on the project files selected by the user. Then, in the workflow corresponding to the approval node, the approval form is pushed to the target approver through a preset application (such as a mini-program, APP, etc.).

[0109] Furthermore, traditional archival document collection mainly relies on manual notification and collection, which results in slow document transmission speed and easy loss. The lack of a unified online collection channel makes it difficult for archivists to quickly and comprehensively obtain the required documents, leading to low document collection efficiency.

[0110] To address this technical issue, this embodiment fully utilizes the constructed table, which allows for dynamic adjustment of the project schedule according to different project needs. The table's additional embedded project process nodes, required file types, and uploaded file cells enable the management and collection of power project archives, thereby improving file collection efficiency.

[0111] Specifically, in this embodiment, when collecting and downloading project files, the project files and file sources of each target project file can be determined based on a table. Then, the project files can be automatically collected and downloaded based on the file sources.

[0112] In practical implementation, after collecting project files, they can be initially organized, such as by initially classifying the uploaded project files using a table. The table is updated in real time with the tags and tag names of the uploaded project files, thus updating the category of the uploaded project files in the table. Therefore, this embodiment utilizes a table to unify the collection channels for project files, achieving rapid and comprehensive acquisition of the required project files. The updated table already includes relevant information about the uploaded project files, so it can be compared with the categories output by the model (this will not be detailed in this embodiment).

[0113] Step 320: Extract labels and training data from the table to form a training dataset.

[0114] Step 330: Establish a deep semantic model and train the deep semantic model using the training dataset to obtain a trained deep semantic model.

[0115] In the specific implementation, the training dataset is input into the deep semantic learning text classification model for model training. This includes training the model to extract filenames from input project files, perform word segmentation on filenames, build indexes, perform BERT inference, and extract and classify text features. The final result is a well-trained deep semantic model.

[0116] Therefore, this embodiment uses existing tables to automatically translate them into training data, which facilitates training set collection and enables efficient training set acquisition and model training.

[0117] Step 340: The uploaded project files are processed for file verification and classification using the trained deep semantic model to obtain the file verification results.

[0118] The file verification result includes the category label of the uploaded project file output by the deep semantic model.

[0119] In related technologies, archival storage largely relies on manual classification and storage, lacking intelligent storage management systems. This easily leads to problems such as chaotic file storage locations, difficulty in retrieval, and an increased risk of file loss or damage. Manual file uploads are prone to errors, such as sending files to the wrong location. Secondly, the sheer volume and disorganized classification of files are a significant issue: with business growth, the number of archival documents has exploded, resulting in massive piles of files. Simultaneously, file classification standards are unclear and inconsistent, with different departments or personnel having varying understandings and operational methods for file classification, leading to chaotic and disorganized file categorization. This makes finding specific files akin to searching for a needle in a haystack, consuming considerable time and effort and severely impacting the efficiency and quality of archival management.

[0120] To address the aforementioned technical issues, this embodiment employs a verification and correction method based on a pre-trained deep semantic learning model to achieve automated verification of uploaded archive files.

[0121] Specifically, the uploaded project files are input into a pre-trained deep semantic model, which then performs verification and classification on the project files. The model can perform multiple processing steps on the input uploaded project files, including extracting file information and further extracting text features. Through text feature analysis, the model can predict and output the category to which the uploaded project file belongs, such as outputting the category with the highest probability, thus obtaining the file verification result.

[0122] Optionally, the above-mentioned file verification and classification processing of uploaded project files using a trained deep semantic model to obtain file verification results may include the following sub-steps:

[0123] Sub-step 3401: In the deep semantic model, the uploaded project file is identified and extracted based on the input to obtain the file name.

[0124] Sub-step 3402 involves processing the file name through word segmentation mapping to construct vector information.

[0125] The vector information includes BERT word vectors.

[0126] Sub-step 3403: Perform feature analysis based on the vector information to obtain text features.

[0127] Sub-step 3404: Use the text features to perform classification verification to obtain the file verification result.

[0128] A unified description is provided for sub-steps 3401-3404:

[0129] In its implementation, the deep semantic model can identify and extract file information such as the filename (or simply file name) from the input project files. Alternatively, other methods can be used to extract the filename of uploaded project files and input into the model; this embodiment does not impose any restrictions on this. Considering that the filename of some project files may not directly reflect the category to which the project file belongs, the deep semantic model can identify the category of the project file by recognizing its content.

[0130] In practical implementation, refer to Figure 7 Taking extracting filenames from uploaded project files as an example, the filenames can be segmented and mapped to obtain multiple segments or words, constructing a word list. Then, each word in the word list can be indexed and mapped to create a corresponding index, forming an index list. Subsequently, BERT (Bidirectional Encoder Representations from Transformers) inference is used to convert the index list data into BERT word vectors to extract text features. Finally, the text features are concatenated and the model outputs the category with the highest probability, i.e., the category label. For example, this can be combined with... Figure 4 As shown in Table 2, if the project file name is "XXX project approval document", the category label predicted by the model will be "11".

[0131] In an optional embodiment, the above-described processing of the file name through word segmentation mapping to construct vector information may specifically include: processing the file name through word segmentation mapping to obtain a word list, the word list including at least one text segment extracted from the file name; performing index mapping processing based on each text segment to establish an index list; and performing BERT inference based on the index list to obtain vector information.

[0132] Step 350: For the uploaded project file, compare the category tag with the tag corresponding to the uploaded project file in the table to obtain the comparison result.

[0133] In the implementation, the table has already been updated with information about the uploaded project files, so the table already contains the categories, i.e., tags, of the uploaded project files. After obtaining the category tags output by the model, the comparison of whether the categories are correct can begin. For example, for "XXX Project Proposal Document", the category tag output by the model is "11". In the table, the corresponding name is found through the training data column, and the tag is found based on the row where the corresponding name is located. Assuming that the tag of "XXX Project Proposal Document" in the table is 11, the category tag is consistent with the tag in the table, and a consistent comparison result is obtained, which is taken as the comparison result. Assuming that the tag of "XXX Project Proposal Document" in the table is 21, the category tag is inconsistent with the tag in the table, and an inconsistent comparison result is obtained, which is taken as the comparison result.

[0134] Step 360: Based on the comparison results, output correction reminder information.

[0135] In one optional embodiment, outputting correction reminder information based on the comparison result may specifically include: when the comparison result is an inconsistency result, obtaining the first tag name of the category label in the table; obtaining the second tag name of the uploaded project file currently in the table; generating correction reminder information based on the first tag name and the second tag name, and outputting the correction reminder information.

[0136] In practice, if the comparison result is consistent, it indicates that the category of the uploaded project file is correct and its corresponding upload location can be considered correct, so no reminder is needed; if the comparison result is inconsistent, it indicates that there is a dispute over the category of the uploaded project file and its corresponding upload location may be incorrect, so a reminder needs to be output.

[0137] Specifically, based on the category labels output by the model, the corresponding label is searched in the table. Then, based on the found label, the label name is further searched, and this searched label name serves as the first label name. Next, since the table already stores information about the uploaded project file, the label name stored in the table is obtained, which is the second label name. Using both the first and second label names, a correction reminder message is generated, which includes the file cell location of the uploaded project file.

[0138] Therefore, this embodiment realizes the management and control of power project archives, and verifies whether the corresponding file cell has been uploaded by file name, and provides a reminder of the corresponding file cell location when an incorrect upload is made.

[0139] Furthermore, in actual implementation, after outputting category reminder information, the project files that may have been uploaded incorrectly can be re-uploaded to the correct file cell based on the category of the table or the category label output by the model. This embodiment does not impose any restrictions on this.

[0140] In practice, users can identify incorrectly uploaded project files based on category alerts and manually re-upload them to their correct target project archive.

[0141] Alternatively, for incorrectly uploaded project files, the uploaded project files can be automatically re-uploaded based on the category tags. Specifically, based on the category tags output by the model, the actual target project file to which the uploaded project file belongs can be determined, the file can be archived under the target project, and then the project file can be automatically re-uploaded to the correct target location.

[0142] As can be seen, this embodiment uses a deep semantic learning classification model for the classification and verification of archive files, which can quickly and efficiently verify and correct a large number of project files, reduce the error rate of file archiving, and improve the efficiency of project management.

[0143] Therefore, this embodiment constructs a comprehensive system covering multiple core modules, including document collection, approval workflow, and one-click download and automatic transfer. Its core objective is to significantly improve the standardization and overall efficiency of document archiving. In traditional archiving processes, project archives consist of a large number of documents with disorganized classifications. Each collection requires approval of each document, which is not only inefficient but also prone to errors and omissions, posing a significant challenge to management. This system, however, utilizes an e-reporting platform to achieve end-to-end digital management from collection and approval to archiving. Specifically, when a user uploads project attachments, the system can quickly and accurately match preset approval rules and material lists, and push this information to the corresponding approval nodes in real time. This ensures the completeness and compliance of archived materials, greatly improving the accuracy and efficiency of approvals.

[0144] In summary, this embodiment, driven by intelligent processes, not only achieves standardization and intelligentization of archival work, ensuring transparency and accuracy in archival management, but also significantly improves the standardization level and efficiency of archival management. Specifically, in the document collection stage, it achieves automatic capture and preliminary classification of documents from multiple channels and in multiple formats, greatly reducing the burden of manual processing and lowering the human error rate. Secondly, for the verification and correction of large batches of project documents, this embodiment utilizes tables to quickly translate training sets, facilitating the collection of training sets. The training sets can be directly used to train a deep semantic learning classification model, enabling the classification and verification of archival documents using the deep semantic learning classification model, effectively reducing archival management risks and improving project archiving efficiency.

[0145] Furthermore, in terms of adjudication, this embodiment has achieved online and standardized processing of the archived materials approval process, with real-time tracking of approval progress and automatic recording of approval opinions. This not only speeds up the approval process but also enhances its transparency and traceability. Regarding cost control, by reducing the use of paper archives, lowering manual processing costs, and improving work efficiency, it saves the organization considerable operating costs.

[0146] As can be seen, this embodiment promotes collaboration between the records management department and other business departments, breaks down information barriers, and realizes the sharing and efficient utilization of records resources. Its standardized and modular design makes the system easy to replicate and promote in other organizations or projects, providing a powerful tool for improving the overall level of records management.

[0147] Therefore, this application focuses on optimizing the archival filing process, aiming to achieve refined control over the entire process of document collection, verification and correction management, and approval of archived materials through digital means. It covers core functions such as document collection, approval workflow, one-click download, and automatic transfer, and is dedicated to solving pain points in traditional archiving work such as the large number of documents, disorganized classification, and the susceptibility to errors and omissions in manual approval, significantly reducing the difficulty of archival management and improving overall work efficiency.

[0148] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should know that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps may be performed in other orders or simultaneously.

[0149] like Figure 8 As shown in the illustration, this application also provides a project file verification and correction system 800 based on a table-based approach throughout the project lifecycle, comprising:

[0150] The table creation module 810 is used to create tables in the established adaptive e-reporting platform, based on the project archives throughout the entire project lifecycle and the complete historical data uploaded by different project types.

[0151] The training dataset extraction module 820 is used to extract labels and training data from the table to form a training dataset.

[0152] The model training module 830 is used to establish a deep semantic model, and to train the deep semantic model using the training dataset to obtain a trained deep semantic model.

[0153] The verification and correction module 840 is used to perform file verification and classification processing on the uploaded project files through a trained deep semantic model, and to compare the file verification results with the categories in the table to obtain the comparison results.

[0154] The correction reminder module 850 is used to output correction reminder information based on the comparison results.

[0155] Optional, the table creation module includes:

[0156] The data identification submodule is used to identify the archived data column and attachment column from the complete historical upload data of different project types;

[0157] The conversion submodule is used to convert the archived data column into tag data and tag name data, and to extract attachment name data from the attachment column;

[0158] The table creation and update submodule is used to create a table and update the label column, label name column, and data name column in the table based on the label data, the label name data, and the attachment name data, respectively. In the table, a mapping relationship is established between each data point in the label column, each data point in the label name column, and each data point in the data name column, and the table is updated through this mapping relationship. Specifically, each data point in the label column serves as a label, each data point in the data name column serves as training data, and each data point in the label name column serves as a label name for correction reminders.

[0159] Optionally, the table creation and update submodule is specifically used to transform the tag data according to a preset transformation rule to obtain the target tag; and to update the tag column in the table based on the target tag.

[0160] Optional, the verification and correction module includes:

[0161] The verification and classification submodule is used to perform file verification and classification processing on the uploaded project files using a trained deep semantic model, and obtain file verification results. The file verification results include the category labels of the uploaded project files output by the deep semantic model.

[0162] The category comparison submodule is used to compare the category label with the corresponding label of the uploaded project file in the table for the uploaded project file, and obtain the comparison result.

[0163] Optional, the verification and classification submodule includes:

[0164] The identification unit is used to perform identification and extraction processing on the uploaded project file based on the input in the deep semantic model to obtain the file name;

[0165] The word segmentation mapping unit is used to process the file name through word segmentation mapping and construct vector information, the vector information including BERT word vectors;

[0166] The feature analysis unit is used to perform feature analysis based on the vector information to obtain text features;

[0167] The classification prediction unit is used to perform classification verification using the text features to obtain the document verification result.

[0168] Optionally, the word segmentation mapping unit is specifically used to process the file name through word segmentation mapping to obtain a word list, the word list including at least one text segment extracted from the file name; perform index mapping processing based on each text segment to establish an index list; and perform BERT inference based on the index list to obtain vector information.

[0169] Optionally, the correction reminder module is specifically used to, when the comparison result is an inconsistency result, obtain the first tag name of the category tag in the table; obtain the second tag name of the uploaded project file in the table; generate correction reminder information based on the first tag name and the second tag name, and output the correction reminder information.

[0170] Optionally, a project file verification and correction system based on the entire project lifecycle and tables may also include:

[0171] The one-click download module is used to extract the target attachment name and category content from the download request of the target object; based on the target attachment name and category content, it traverses the table and performs hierarchical retrieval using a table organization method to obtain the file data to be downloaded; based on the file data to be downloaded, it recursively adds files to construct a compressed package; and downloads the compressed package using a preset download tool.

[0172] It should be noted that the project file verification and correction system based on a project lifecycle combined with a table provided in this application embodiment can execute the project file verification and correction method based on a project lifecycle combined with a table provided in any embodiment of this application, and has the corresponding functions and beneficial effects of the execution method.

[0173] In practical implementation, the aforementioned project file verification and correction system based on a table-based approach throughout the project lifecycle can be integrated into a device. This device can quickly translate training sets based on tables, train models using the training sets, and then rapidly verify and correct large batches of project files using the models. As an electronic device, it automates the organization of project files, significantly reducing the difficulty of file management and improving overall work efficiency. This electronic device can consist of two or more physical entities, or it can consist of a single physical entity. For example, the electronic device can be a personal computer (PC), a computer, a server, etc. This application embodiment does not impose specific limitations in this regard.

[0174] likeFigure 9 As shown, this application embodiment provides an electronic device, including a processor 111, a communication interface 112, a memory 113, and a communication bus 114. The processor 111, communication interface 112, and memory 113 communicate with each other via the communication bus 114. The memory 113 stores computer programs. When the processor 111 executes the program stored in the memory 113, it implements the steps of a project file verification and correction method based on a table combined with the entire project lifecycle, as provided in any of the aforementioned method embodiments. For example, this method may include the following steps: In an established adaptive e-report platform, based on the project files throughout the entire project lifecycle, a table is established using historical uploaded complete data of different project types; tags and training data are extracted from the table to form a training dataset; a deep semantic model is established, and the deep semantic model is trained using the training dataset to obtain a trained deep semantic model; the uploaded project files are processed for file verification and classification using the trained deep semantic model, and a comparison is performed between the file verification results and the categories in the table to obtain a comparison result; based on the comparison result, a correction reminder message is output.

[0175] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0176] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for verifying and correcting project archives based on a table-based approach throughout the project lifecycle, characterized in that: include: In the established adaptive e-reporting platform, based on the establishment of project archives for the entire project lifecycle, tables are created based on the historical uploaded complete data of different project types. Extract labels and training data from the table to form a training dataset; Establish a deep semantic model, and train the deep semantic model using the training dataset to obtain a trained deep semantic model. The uploaded project files are verified and classified using a trained deep semantic model. The file verification results are then compared with the categories in the table to obtain the comparison results. Based on the comparison results, a correction reminder message is output.

2. The method according to claim 1, characterized in that, Create tables based on the complete historical upload data for different project types, including: Identify the archived data column and attachment column from the complete historical upload data of different project types; Convert the archived data column into tag data and tag name data, and extract the attachment name data from the attachment column; Create a table, and update the label column in the table based on the label data, update the label name column in the table based on the label name data, and update the data name column in the table based on the attachment name data, respectively; In the table, a mapping relationship is established between each data in the label column, each data in the label name column, and each data in the data name column, and the table is updated through the mapping relationship; In this table, each data point in the label column serves as a label, each data point in the data name column serves as training data, and each data point in the label name column serves as a label name used for correction reminders.

3. The method according to claim 2, characterized in that, Updating the label column in the table based on the label data includes: The tag data is converted according to a preset conversion rule to obtain the target tag; Update the label column in the table based on the target label.

4. The method according to claim 1, characterized in that, The uploaded project files are processed for file verification and classification using a trained deep semantic model. The file verification results are then compared with the categories in the table to obtain the comparison results, including: The uploaded project files are processed by a trained deep semantic model to perform file verification and classification, and the file verification results include the category labels of the uploaded project files output by the deep semantic model. For the uploaded project file, the category tag is compared with the tag corresponding to the uploaded project file in the table to obtain the comparison result.

5. The method according to claim 4, characterized in that, The uploaded project files are processed for file verification and classification using a trained deep semantic model to obtain the file verification results, including: In the deep semantic model, the uploaded project file is identified and extracted based on the input to obtain the file name; The file name is processed by word segmentation mapping to construct vector information, which includes BERT word vectors. Based on the vector information, feature analysis is performed to obtain text features; The text features are used for classification and verification to obtain the file verification result.

6. The method according to claim 5, characterized in that, The file names are processed through word segmentation mapping to construct vector information, including: The file name is processed by word segmentation mapping to obtain a word list, which includes at least one text word extracted from the file name; Based on the text segmentation described above, an index mapping process is performed to establish an index list; Based on the index list, BERT inference is performed to obtain vector information.

7. The method according to claim 4, characterized in that, Based on the comparison results, a correction reminder message is output, including: When the comparison result is inconsistent, obtain the first label name of the category label in the table; Get the name of the second tag of the uploaded project file in the table; Based on the first tag name and the second tag name, a correction reminder message is generated and output.

8. The method according to any one of claims 1-7, characterized in that, After creating the table, it also includes: Extract the target attachment name and category from the download request of the target object; Based on the target attachment name and the category content, the table is traversed, and a table organization method is used to perform hierarchical retrieval to obtain the file data to be downloaded; Based on the data of the files to be downloaded, a compressed package is constructed by recursively adding files; Download the compressed package using the preset download tool.

9. A project file verification and correction system based on a table-based approach throughout the project lifecycle, characterized in that: include: The table creation module is used to create tables in the established adaptive e-reporting platform, based on the project archives throughout the entire project lifecycle and the complete historical data uploaded for different project types. The training dataset extraction module is used to extract labels and training data from the table to form a training dataset. The model training module is used to build a deep semantic model, and to train the deep semantic model using the training dataset to obtain a trained deep semantic model. The verification and correction module is used to perform file verification and classification processing on uploaded project files through a trained deep semantic model, and to compare the file verification results with the categories in the table to obtain the comparison results. The correction reminder module is used to output correction reminder information based on the comparison results.

10. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements the steps of the project file verification and correction method based on a table combining the entire project lifecycle as described in any one of claims 1-8.

Images